Case Study:  Using the Geomedix™ Process to Improve Clinical Trial Outcomes

Background

A pharmaceuticals manufacturer was conducting a Phase 3 study of a new drug formulated to help control arteriosclerosis. Based on their experience in earlier phases of this study, the manufacturer and their clinical trial management organization selected four cities (Atlanta, Baltimore, Cincinnati, and Detroit) and one investigative site per city (A1, B1, C1, and D1, respectively). Each site had been given a patient-recruiting quota of 50 participants. After 12 months of their 18-month campaign, a total of 150 patients had been recruited. The results by site:

The trial’s sponsor pondered the factors that might have affected the trial’s recruiting efforts. The campaign messaging might not have been relevant. The media selection might not have been on target. Or the entire campaign may have been under-funded.

But another important factor, and one that can be measured, was proximity of the sites to qualified patients.

For that reason, the team decided to utilize the Geomedix™ process to evaluate the patient population proximate to the trial’s investigative sites.

Market Evaluation

Geomedix employs a sophisticated model that considers all of the known characteristics of a medical condition, weights each factor in terms of its relevance to the diagnosis, and then estimates the percentage of a population likely to have been diagnosed with the condition.

There are many statistically meaningful indicators associated with arteriosclerosis.  Prevalence and incidence figures are available and there are several demographic indicators that are strongly linked to the disease. The Geomedix model also considers risk factors, physician data, prescription reporting and other medical data to refine its population estimates. 

So as a first step, all of these factors were loaded into the Geomedix model in order to examine prevalence within media markets. The ensuing analysis produced the following data.

Top Arteriosclerosis Markets

Sample data.

This analysis reveals two important findings—the available pool of qualified patients, and the comparative media efficiency of each market. Thus the table creates a comparison of media costs to patient populations, the result of which are the attendant disease media efficiency values.

Had the sponsor completed this analysis prior to the original campaign execution, he would have discovered that Detroit was number 19th ($27.81/K) while Cincinnati ranked 32nd ($34.98/K).

Site Evaluation

The manufacturer then decided to extend the analysis to the investigative site level by first selecting two alternate sites in the two most media-efficient cities, Atlanta and Baltimore, and then to use the results of that analysis to compare site performance in two new media-efficient cities, Tucson and Ft. Lauderdale. Two sites in Detroit and Cincinnati were kept as control points.

To make the comparisons transportation-neutral, a drive-time (rather than mileage) band was placed around each site in each city. The sponsor felt that a 20-minute drive time was the limit for weekly trial participation. Anything more would negatively impact both recruiting and retention. So in Atlanta, for example, that translated into a 12-mile radius around each site.

Investigative Site Recruiting Potential--Basis: 20-minute bands

Sample data.

The above site comparisons clearly show why Atlanta and Baltimore outperformed Cincinnati and Detroit. Perhaps other sites in both of the latter markets might have fared better, but with such low market efficiency, why bother to find out?

Atlanta’s A1 investigative site was the only to reach 100% of quota. In that case, it took a reference population of 3,817 to produce 50 patients. Baltimore’s site, B1, produced 40 patients with an available pool of 3,238. From these data points we might infer that it takes roughly 75 persons within a site radius to produce one qualified patient. That would reduce our qualification criteria to a site that had at least 3,800 potential patients.[3]

The difference between Atlanta’s sites A1 and A3 is negligible, certainly within the margin of error associated with the Geomedix process. Baltimore’s B1 and B2 strike a much greater differential, and the sponsor would do well to consider the second site as an adjunct to the first. Ft. Lauderdale’s F1 and Tucson’s T3 are clearly the most suitable candidates in those markets.

Summary

By conducting both the market efficiency analysis as well as the site banding evaluation, pharmaceutical manufacturers and their trial management organizations can perform two very important evaluations: In the case where quotas were filled early in all locations, how much less could they have spent and still met expectations? And in the case where quotas were not met, what markets would have produced denser patient populations for comparable or even reduced media costs?

The Geomedix process is a robust analytical tool that can provide the answers to both questions, thereby assuring greater return on investment in almost all clinical trial recruitment campaigns.

[1] The Media Cost is a proportionally weighted value associated with the media mix that the sponsor or its trial management organization is utilizing.

[2] Expected Recruits is the Disease Population divided by 75 (see discussion).

[3] We are, of course, ignoring the time factor here. It may have turned out that the Atlanta investigative site achieved their quota in only two months while the others ran the entire campaign period and still failed to meet their objectives.  A more in-depth analysis would likely compare recruitment prevalence over time to make a completely realistic comparison.

Top of Page

For more information, contact us at 770-931-9677 or email at info@pkdata.net